Cover Page The handle

The handle http://hdl.handle.net/1887/66032 holds various files of this Leiden University

dissertation.

Author: Fleurbaaij, F.

Title: Novel applications of mass spectrometry-based proteomics in clinical microbiology

Issue Date: 2018-09-27

CHAPTER 2

Capillary-electrophoresis mass spectrometry for

the detection of carbapenemases in (multi-)

drug-resistant Gram-negative bacteria

Frank Fleurbaaij¹, Anthonius A.M. Heemskerk², Anne Russcher¹, Oleg I. Klychnikov², André M. Deelder², Oleg A. Mayboroda², Ed J. Kuijper¹, Hans C. van Leeuwen^1#, Paul J.

Hensbergen^2#

Analytical Chemistry, 2014, 86 (18), pp 9154–9161

¹ Department of Medical Microbiology, Section Experimental Microbiology, Leiden University Medical Center, Leiden, the Netherlands

2Center for Proteomics and Metabolomics, Leiden University Medical Center, Leiden, the Netherlands

# These authors contributed equally

Abstract

In a time in which the spread of multi-drug resistant microorganisms is ever increasing, there is a need for fast and unequivocal identification of suspect organisms to supplement existing techniques in the clinical laboratory, especially in single bacterial colonies. Mass- spectrometry coupled with efficient peptide separation techniques offer great potential for identification of resistant-related proteins in complex microbiological samples in an unbiased manner.

Here we developed a capillary electrophoresis-electrospray ionization-tandem mass spectrometry CE-ESI-MS/MS bottom-up proteomics workflow for sensitive and specific peptide analysis with the emphasis on the identification of β-lactamases (carbapenemases OXA-48 and KPC in particular) in bacterial species. For this purpose, tryptic peptides from whole cell lysates were analysed by sheathless CE-ESI-MS/MS and proteins were identified after searching of the spectral data against bacterial protein databases.

The CE-ESI-MS/MS workflow was first evaluated using a recombinant TEM-1 β-lactamase, resulting in 68% of the amino acid sequence being covered by 20 different unique peptides.

Subsequently, a resistant and susceptible Escherichia coli lab strain were analysed and based on the observed β-lactamase peptides, the two strains could easily be discriminated. Finally, the method was tested in an unbiased setup using a collection of in-house characterised OXA-48 (n=17) and KPC (n=10) clinical isolates. The developed CE-ESI-MS/MS method was able to identify the presence of OXA-48 and KPC in all of the carbapenemase positive samples, independent of species and degree of susceptibility. Four negative controls were tested and classified as negative by this method. Furthermore, a number of extended-spectrum beta- lactamases (ESBL) were identified in the same analyses, confirming the multi-resistant character in 19 out of 27 clinical isolates. Importantly, the method performed equally well on protein lysates from single colonies. As such, it demonstrates CE-ESI-MS/MS as a potential next generation mass spectrometry platform within the clinical microbiology laboratory.

Chapter 2 Introduction

Antibiotic resistance is a growing problem of modern medicine. Micro-organisms that are resistant to a range of antibiotics are becoming widespread and at the same time the development of novel antibiotics is slowing down¹. One of the ways to respond to the increase of multidrug resistant microorganisms is to ensure a fast identification of suspect organisms to allow for appropriate infection prevention strategies. As an example, for hospital-acquired methicillin-resistant Staphylococcus aureus (MRSA) appropriate screening methods have proven to be a key factor in the decline in prevalence in recent years².

One of the most important antibiotic classes with a broad range of antimicrobial activity is the β-lactam group, all sharing a structural similarity in possessing the β-lactam ring.

They act by inhibiting enzymes involved in cell wall synthesis which eventually results in cell death³. They are widely applied and as a result there has been an increase in resistance to this class of antibiotics. Resistance to β-lactam antibiotics is predominantly caused by the expression of β-lactamases, which hydrolyse the β-lactam ring rendering the antibiotic inactive⁴. The carbapenem class of antibiotics is a keystone in the treatment of bacteria that have developed a resistance against other type of β-lactam antibiotics through extended- spectrum β-lactamases (ESBLs)⁵. Carbapenem resistance in turn is now developing and becoming more widespread. The most prevalent carbapenemases are KPC, VIM, IMP, NDM-1 and OXA-48⁶. OXA-48 and Klebsiella pneumoniae carbapenemases (KPC) are two important carbapenemases that are found in various bacterial species⁷. OXA-48 has first been identified in Turkey in 2001 and has proliferated ever since. Recent years have seen outbreaks of OXA-48 positive bacteria in a number of countries⁸.

The proteomic fingerprinting using MALDI-ToF MS has been implemented in many medical microbiological laboratories as a tool for the identification of bacteria⁹. However, this approach is usually insufficient for direct identification of resistant phenotypes and is not suited for confident identification of the specific proteins accountable for this resistance. For this purpose, true identification of proteins involved in resistance, such as beta-lactamases, is necessary. Bottom-up proteomics analysis using liquid chromatography coupled to mass spectrometry is a very well accepted approach for this kind of analyses and has recently also been used to study β-lactam resistance in Acinetobacter baumannii¹⁰. An alternative technical solution could be capillary electrophoresis (CE). CE is a highly efficient method for peptide analysis since it is able to separate and analyse peptides with a wide range of physico-chemical properties, and compared to liquid chromatography-mass spectrometry (LC-MS), is particularly suited for the analysis of smaller peptides¹¹. The recently developed innovative interfaces (both employing sheath-liquid and sheathless variants) have considerably advanced the coupling of CE with MS with the key benefit of the possibility to work at ultra-low flow rates, resulting in reduced ion suppression and improved sensitivity^12,13.

The analytical parameters and advantages of such methods for peptide analysis have recently been evaluated^14,15.

All these aspects indicate that CE-MS is a potentially valuable technique for confident identification of, and differentiation between, specific beta-lactamases in bacteria. However, the real application and performance of such a method in the analyses of clinically relevant samples is still lacking. The aim of this study was to develop a CE-ESI-MS/MS based method as well as compatible sample preparation for the identification of beta-lactamases in complex samples. Because of the difficulty with current methodologies to identify OXA-48 and KPC producing Enterobacteriaceae¹⁶, we decided to test the method on a well-defined set of such clinical isolates.

Materials and methods

Characterization of the reference set of Gram-negative clinical isolates

The collection of Gram-negative clinical isolates (Klebsiella pneumoniae, Escherichia coli and Enterococcus cloacae) consisted of isolates obtained in-house from patients admitted to the hospital as well as a number of isolates from a reference set of the Dutch National Institute for Public Health and Environment (RIVM, Bilthoven, the Netherlands). The meropenem minimum inhibitory concentration (MIC) was determined using an E-test according to manufacturer’s instructions (Biomérieux Benelux, Zaltbommel, the Netherlands). Screening for carbapenemase production was performed by a modified Hodge-test, as previously described by Lee¹⁷. All isolates were characterized at the molecular level; most OXA-48 isolates were characterized by a PCR assay at the RIVM. One was characterized by PCR at the carbapenemase reference laboratory of P. Nordmann (South-Paris Medical School, Paris, France) and one was characterized by MicroArray (Checkpoints, Wageningen 4, the Netherlands). All KPC isolates were characterized by a real-time in-house PCR assay at the LUMC.

MALDI-ToF MS analysis of carbapenem breakdown was performed according to the protocol as described by Sparbier¹⁸ with slight modifications. In short: bacteria were cultured for 18-24 hours on Columbia blood agar plates (Biomerieux Benelux, Zaltbommel, the Netherlands).

The amount of bacteria filling an 1 μl inoculation loop were suspended in 30 μl ertapenem (Invanz®, Merck, Haarlem, the Netherlands), 0.5 mg/ml in 10 mM ammonium hydrogen citrate (AHC), pH 7) and incubated for 3 hours. After centrifugation for 2 min at 13.000 x g, 1 μL of the cell-free supernatant was spotted onto a steel polished MALDI plate in triplicate and dried at room temperature. Subsequently the spot was covered with 1 μl α-HCCA in 50% acetonitrile-2.5% trifluoroacetic acid. Each run included a negative control (E. coli ATCC 25922 incubated in the presence of ertapenem) and an ertapenem control (ertapenem in

Chapter 2 AHC), both pre-treated as above. Fifteen carbapenemase negative strains from the American

Type Culture Collection (ATCC) were also tested.

Spectra were obtained with a Microflex LT mass spectrometer (Bruker Daltonics, Bremen, Germany). The spectrum per spot was the product of 240 shots. Spectra were measured in positive linear mode in a mass range van 100 to 1100 Da. Spectra were analysed using FlexAnalysis 3.3 (Bruker Daltonics) in the range from m/z 440 to 580. Successful identification of carbapenemase activity was defined as complete disappearance of peaks at m/z 475.5, m/z 498.5 and m/z 520.519.

Cell culturing, lysis and in-solution protein digestion using trypsin

Cells were grown in Brain Heart Infusion broth (Biomérieux Benelux, Zaltbommel, the Netherlands) and pellets from 1 ml cultures after 1 min 10.000g were re-suspended in a fixed volume of deionized water. Total protein content in these samples was analysed with the Pierce Micro BCA Protein Assay kit (Thermo Scientific), following the manufacturer’s instructions. The samples were then stored at -80 °C until further use.

For the lysis and digestion of the samples an adapted protocol based on the protocol by Wang et al.²⁰ was used. In short, a cell suspension corresponding to 20 µg of protein content was diluted 1:1 (v/v) in trifluoroethanol (TFE). The sample was vortexed and incubated at 60 °C for 1 hour. The sample was then sonicated in a water bath for 2 min. Ammonium bicarbonate and dithiothreitol (DTT) were added with final concentrations of 25 mM and 2.5 mM respectively and the sample was reduced at 55 °C for 15 min. For the alkylation, iodoacetamide (in 25 mM ammoniumbicarbonate) was added to final concentrations of 5.5 mM. Alkylation took place in the dark at room temperature for 15 min. Trypsin was added to the sample at a 1:20 enzyme to protein ratio and the sample was digested overnight at 37

°C. The digestion was quenched with 10% acetic acid and the samples were lyophilized and subsequently reconstituted in 1% acetic acid and 50 mM ammonium acetate pH 4.

For the analyses of single bacterial colonies, cells were grown on Columbia blood agar plates as described above. Bacterial colonies were picked individually by re-suspending them in 20 µl of deionized water. Then, 20 µl of TFE was added and lysis and tryspin digestion was subsequently performed as described above.

CE-ESI-MS(/MS) analysis

Capillary-electrophoresis mass spectrometry analysis was performed on a PA800 plus (Beckman Coulter, Brea, CA) coupled to a maXis Impact UHR-TOF-MS (Bruker Daltonics) utilizing an interface based on the initial design by Moini²¹. Prior to analysis, the capillary was sequentially rinsed with 0.1 M NaOH, 0.1 M HCl, deionized water and finally the background electrolyte, which consisted of 10% acetic acid. Samples were hydrodynamically injected at

1 psi for 60 s, which results in a sample plug equivalent to 1% of the total capillary volume.

The separation was performed at 20 kV for 60 min using a bare fused-silica capillary. Eluting peptides were analyzed using the data dependent MS/MS mode over an m/z 300-2000 range.

The 10 most abundant ions in an MS spectrum were selected for MS/MS analysis by collision- induced dissociation using helium as the collision gas. A 1 minute dynamic exclusion window was used for precursor selection and the window for fragment analysis ranged from m/z 300-2200.

Data analysis

A database was prepared using a collection of gram-negative bacteria. The databases were obtained from http://org.uib.no/prokaryotedb/²². This final database was supplemented in- house with various β-lactamase sequences. Peak lists were generated using Data analysis 4.0 (Bruker Daltonics, Bremen, Germany) with default settings and exported as Mascot Generic Files. Peptides were identified using the Mascot algorithm (Mascot 2.4.1, Matrix Science, London, UK) using Mascot Deamon 2.2.2. A MS tolerance of 0.05 Da and a MS/MS tolerance of 0.8 Da were used. Trypsin was designated as the enzyme and up to one missed cleavage site was allowed. Carbamidomethylcysteine was selected as a fixed modification and oxidation of methionine as a variable modification. The false discovery rate (peptide matches above the identity threshold) was set at 1% using a decoy database. For the scoring of beta-lactamases, only beta-lactamase hits with at least two unique peptides with a score above 25 were selected.

Results

CE-ESI-MS(/MS) analysis of recombinant TEM beta-lactamase and ampicillin-resistant and susceptible E. coli laboratory strains

The performance of the method was tested using a tryptic digest of recombinant TEM-1 beta-lactamase. Using 1 ng of protein, TEM-1 was identified with 20 unique peptides and a resulting total sequence coverage of 68% (data not shown). To test the applicability of detecting this beta-lactamase in a complex sample, an ampicillin resistant E. coli lab strain (BL21(DE3)) containing a plasmid encoding a TEM β-lactamase was tested along with the same ampicillin sensitive strain lacking this plasmid. Hence, the only difference between the two strains is the expression of β-lactamase in the resistant strain. Tryptic digests of bacterial lysates were analysed in triplicate by CE-ESI-MS (Figure 1A). Based on the in-silico tryptic digest of TEM ß-lactamase, the CE-ESI-MS runs were screened for the presence or absence of the predicted ß-lactamase derived tryptic peptides as exemplified by several extracted ion chromatograms shown in Fig. 2B and C.

Chapter 2

Figure 1: CESI-MS/MS analysis of a tryptic digest of an ampicillin sensitive and resistant E. coli strain.

A: Representative base peak electropherogram as observed for an E. coli tryptic digest. B and C: Extracted ion chromatograms (EIC) of several m/z values corresponding to tryptic peptides of beta-lactamase that are absent in the ampicillin sensitive strain (B) but present in the resistant strain (C). D and E: Mass spectrum at a fixed time point (30.3 min) showing the β-lactamase peptide DAEDQLGAR (m/z 487.7294) in the resistant strain (E) which does not appear in the sensitive strain (D). The other peaks within the spectrum correspond to non-β-lactamase peptides, shared by the resistant and sensitive strain.

FPMMSTFK (494.735 +/- 0.01)

IDAGQEQLGR (543.781 +/- 0.01) DAEDQLGAR (487.73 +/- 0.01) VGYIELDLNSGK (654.346 +/- 0.01)

WEPELNEAIPNDER (856.40 +/- 0.01)

Ampicillin resistant Ampicillin sensitive

474.7379

476.7170

478.2623

480.2792

483.7394

490.7365

474.7371

478.2620

480.2794

483.7403

487.7294

490.7362

0

1

2

3

4⁴

x10

Intens.

0

2

4

6⁴

x10

474

DAEDQLGAR

Ampicillin sensitive

Ampicillin resistant ß-lactamase peptide

A

C

B

D

E

Table 1 overview of the clinical isolates used in this study. Listed are the carbapenemases (KPC and oXA-48) as identified by PCR, the results of the modified Hodge test, an E-test (Minimal Inhibitory Concentration (MIC)) and the MALDI-ToF MS ertapenem breakdown analysis. In the last column the results of the bottom-up proteomics analysis by CESI-MS/MS are listed showing that all target β-lactamases are identified as well as a number of extended-spectrum β-lactamases. See material and methods for further details.

Carbepenemase (PCR)

Species: Hodge test: Meropenem (MIC in mg/L)

MALDI-ToF MS (Ertapenem degradation)

CESI-MS/MS (bottom-up proteomics)

KPC K. pneumoniae pos >32 ID + KPC

KPC K. pneumoniae pos > 32 ID + KPC

KPC K. pneumoniae pos 2 ID + KPC/SHV

KPC K. pneumoniae pos 2 ID + KPC

KPC K. pneumoniae pos 32 ID + KPC/SHV

KPC K. pneumoniae pos 32 ID + KPC/SHV

KPC K. pneumoniae pos 12 ID + KPC/TEM/SHV

KPC K. pneumoniae pos > 32 ID + KPC/SHV

KPC K. pneumoniae pos 3 ID + KPC

KPC K. pneumoniae pos 6 ID + KPC

OXA-48 K. pneumoniae pos 24 ID - OXA-48/CTX

OXA-48 K. pneumoniae pos 8 ID - OXA-48/CTX/SHV

OXA-48 K. pneumoniae pos 1,5 ID + OXA-48/CTX

OXA-48 K. pneumoniae pos > 32 ID - OXA-48

OXA-48 K. pneumoniae pos 2 ID - OXA-48/TEM

OXA-48 E. coli pos 0,5 ID - OXA-48/CTX

OXA-48 K. pneumoniae pos 1 ID - OXA-48/CTX

OXA-48 K. pneumoniae pos 3 ID - OXA-48/CTX

OXA-48 K. pneumoniae neg 0,4 ID + OXA-48

OXA-48 K. pneumoniae pos 2 ID + OXA-48

OXA-48 K. pneumoniae pos 24 ID - OXA-48/CTX

OXA-48 E. cloacae pos 1,5 ID - OXA-48/CTX

OXA-48 E. cloacae pos 0,5 ID - OXA-48/ampC

OXA-48 K. pneumoniae pos >32 ID - OXA-48/TEM

OXA-48 K. pneumoniae pos 32 ID - OXA-48/CTX/SHV

OXA-48 K. pneumoniae pos 0,5 ID - OXA-48/CTX

OXA-48 K. pneumoniae pos 0,5 ID - OXA-48/CTX

Chapter 2 For example, the tryptic ß-lactamase peptide DAEDQLGAR (m/z 487.729 [M+2H]²⁺) was

examined. In the extracted ion chromatogram of ions with m/z 487.73 ±0.01 a clear peak is visible in the run of the ampicillin resistant strain which is absent in the sensitive strain (Compare figures 1B and C). Close inspection of the underlying MS spectra (Figure 1D and 1E) indeed shows that amongst other peptides that are co-migrating with this peptide, which are identical in both runs, a clear signal at m/z 487.7294 is present in the spectrum derived from the resistant strain, corresponding to the above mentioned peptide DAEDQLGAR (theoretical m/z 487.7305). In total, 14 of such differential peptides with masses corresponding to ß-lactamase tryptic peptides were identified in all CE-ESI-MS runs from the ampicillin resistant strain, while none of these were identified in the non-resistant strain. The nature of the TEM beta-lactamase peptides was confirmed by MS² analysis (data not shown).

Selection and characterization of clinical isolates

To test the performance of our method for the identification of OXA-48 and KPC carbapenemases, a collection of clinical isolates containing such carbapenemases was needed. For this purpose, a set of samples were selected that were PCR positive for either of these carbapenemases (Table 1). For phenotypic characterization, a modified Hodge test for the detection of carbapenemases was used which showed that all isolates, but one, were positive (Table 1). In addition, the meropenem minimal inhibitory concentration (MIC) was determined using an E-test as a measure of the degree of carbapenem resistance. This showed that the set of isolates varied considerably in their degree of resistance to meropenem (Table 1.) Finally, all KPC isolates were positively identified (10/10) using the MALDI-ToF MS ertapenem hydrolysis assay. However, with this assay only 3/17 of OXA-48 isolates were positively identified (Table 1). A tenfold dilution of ertapenem (0.05 mg/ml) using a subset of samples increased detection to 9 out of 13 isolates without loss in specificity (i.e. ertapenem peaks were still present in all 15 carbapenemase negative strains). Overnight incubation affected specificity as only 5/15 negative strains were still correctly identified.

Analysis of clinical isolates using CE-ESI-MS/MS

The collection of clinical isolates was analysed in triplicate using the CE-ESI-MS/MS platform.

Fragmentation spectra were acquired in a data dependent manner and a MASCOT database search was performed to identify peptides based on the recorded spectra. Importantly, for confident identification of specific beta-lactamases subspecies, unique discriminating peptides need to be detected. A phylogenetic tree based on a sequence alignment using a selection of enzymes covering the various classes of beta-lactamases shows that the differences in the primary amino acid sequences within a certain class may be very small (e.g. the KPC group) while for other classes (e.g. the OXA group), differences in the primary structures are more pronounced (Figure 2A). As such, the discrimination between different classes can be made based on multiple different tryptic peptides, while identification of a single type may rely on an individual peptide. In practice, the lactamases were identified

with a minimum of 4 different peptides using a merged database search of triplicate analysis although all individual analyses yielded lactamase identifications as well. In the case of KPC, the class assignment was confidently performed based on several unique peptides (an example shown in Figure 2B) while within the OXA–group, OXA-48 could be unambiguously identified (Figure 2C).

Figure 2

Figure 2: Differentiation between different β-lactamase subgroups and species.

A: Phylogenetic tree based on a sequence alignment of different β-lactamases from different functional subgroups.

For some classes the primary structures can be highly similar (e.g. KPC and CTX) while for others, individual sequences can vary considerably (e.g. OXA). B: Tandem mass spectrum for a unique tryptic peptide from the KPC-group of carbapenemases (FPLCSSFK). C Tandem mass spectrum for a unique tryptic peptide from OXA-48 (SQGVVVLWNENK).

Chapter 2

OXA-48

1 MRVLALSAVFLVASIIGMPAVAKEWQENKSWNAHFTEHKSQGVVVLWNEN 51 KQQGFTNNLKRANQAFLPASTFKIPNSLIALDLGVVKDEHQVFKWDGQTR 101 DIATWNRDHNLITAMKYSVVPVYQEFARQIGEARMSKMLHAFDYGNEDIS 151 GNVDSFWLDGGIRISATEQISFLRKLYHNKLHVSERSQRIVKQAMLTEAN 201 GDYIIRAKTGYSTRIEPKIGWWVGWVELDDNVWFFAMNMDMPTSDGLGLR 251 QAITKEVLKQEKIIP

AA Observed m/z Mr(calculated) Delta(ppm) Score Peptide

30 – 39 628.792 1255.573 -3 45 SWNAHFTEHK

40 – 51 686.865 1371.715 0.4 61 SQGVVVLWNENK

52 – 60 525.272 1048.530 -1 50 QQGFTNNLK

95 – 100 381.680 761.346 -1.4 20 WDGQTR

101 – 107 438.222 874.430 -1.3 37 DIATWNR

108 – 116 521.770 1041.528 -3.1 21 DHNLITAMK 117 – 128 729.377 1456.735 3.1 15 YSVVPVYQEFAR 164 – 174 632.848 1263.682 -1.5 69 ISATEQISFLR

175 – 180 401.731 801.450 -3.4 33 KLYHNK

181 – 186 370.706 739.340 0.4 42 LHVSER

Figure 3

Sample ID Species: Meropenem OXA-48 sequence

MIC mg/L coverage %

223 C1 E. coli 0.5 27

C2 32

C3 21

225 C1 K. pneumoniae 1 25

C2 29

C3 27

235 C1 E. cloacae 1.5 33

C2 34

C3 29

237 C1 K. pneumoniae >32 31

C2 38

C3 38

A

B

Figure 3 CE-ESI-MS/MS identification of oXA-48 in tryptic digests derived from a single colony. Single bacterial colonies were sampled and processed for tryptic digestion, CE-ESI-MS/MS analysis and database searching.

A: Primary sequence OXA-48 and example of results from one CE-ESI-MS/MS analysis, showing multiple unique OXA- 48 peptides with an overall sequence coverage (in red) of 33%. AA, amino acid number within the primary sequence.

B: CE-ESI-MS/MS identification of OXA-48 in three individual colonies (C1-C3) from different OXA-48 positive bacterial species varying in the degree of susceptibility to meropenem.

Overall, using our CE-ESI-MS/MS platform, the presence of OXA-48 and KPC β-lactamases was demonstrated for all samples measured in triplicate (Table 1), as well as each analysis individually. This demonstrates that this method is capable to phenotypically identify OXA-48 and KPC positive clinical isolates. It is well known that within such isolates multiple different β-lactamases can be present which are often related to the multi-drug resistant phenotype of

such strains. An advantage of our method is that it is unbiased in nature and therefore may also reveal the presence of such resistance related enzymes. Indeed, within our set of clinical isolates a number of extended-spectrum beta-lactamases (ESBLs) such as CTX and SHV were observed (Table 1). 19 out of 27 of the clinical isolates were shown to harbour additional ESBLs as well as the known carbapenemases. A number of negative controls (n=4) were also analysed, in triplicate. These were all negative for beta-lactamases.

Analysis of single bacterial colonies

The applicability of our workflow would benefit from detection on colony level. Therefore, we tested the performance of our method on single colonies after growing on solid medium.

Four isolates from the collection of OXA-48 positive strains were selected to be grown on blood agar plates overnight. From each plate three colonies were sampled and processed independently. Each colony was analysed in triplicate (4 ng of protein injected). Figure 3A shows the results obtained from one CE-ESI-MS/MS analysis of a colony from an OXA- 48 clinical isolate, clearly demonstrating multiple OXA-48 unique peptides. Importantly, in all individual CE-ESI-MS/MS analyses of the independently sampled colonies OXA-48 was identified (Figure 3B). In addition, in several samples ESBLs were identified (data not shown).

Overall, the OXA-48 sequence coverage varied between 21-38% in the different colonies.

Discussion

From the very early days of capillary electrophoresis the potential of the method for highly efficient separation of peptides and proteins was commonly recognised^23-25. Moreover, as a miniaturized technique operating at very low flow rates, CE represents a natural match for the ESI process providing advantages such as reduced ion suppression and improved sensitivity^12,26. Yet, with the emergence of proteomics, liquid chromatography – mass spectrometry (LC-MS) became the mainstream method which pushed CE aside to a position of an interesting but exotic technique. A complex (in comparison to a RPLC-MS) interfacing between CE and MS partially explains the limited use of CE in proteomics. For instance, the most widely used interface, namely a sheath-liquid interface, is a technical compromise which provides a robust CE-MS hyphenation at the expense of sensitivity²⁷. Recent years have seen a push into the development of novel CE-MS interfaces. As much as they are different in the technical details all those interfaces are being developed for a common purpose of maximising the natural advantages of the CE-MS as an ultra-low flow technique.

Here, we present a CE-ESI-MS/MS based workflow for the identification of beta-lactamases in complex samples and demonstrate its applicability for the confident and sensitive analysis of carbapenemases (OXA-48 and KPC) in clinical isolates of Enterobacteriaceae. A sheathless porous sprayer interface was used to couple CE to the mass spectrometer, but we

Chapter 2 are convinced that our approach can easily be translated to the set-ups using other CE-MS

interfaces^28-30.

KPC is the most common carbapenemase found in the United States and has spread worldwide since its discovery in 1996, even becoming endemic in certain countries (e.g.

Israel and Greece)³¹. KPC can be detected by the modified Hodge test and other phenotypic tests but these methods lack speed and can be difficult to interpret. OXA-48 also has become widespread worldwide but detection can be challenging. OXA-48-like producers exhibit a range of susceptibility patterns, and often show only decreased susceptibility to carbapenems, which renders standard susceptibility testing an insufficient screen. As opposed to other carbapenemase genes whose expression can be inhibited in vitro with certain compounds (e.g. EDTA, boronic acid), no phenotypic test can detect OXA-48-like producers in general. The modified Hodge test shows adequate performance³² but requires an overnight incubation step and provides no information about the kind of carbapenemase that is present, which might prove essential in an outbreak-setting. Multiple PCR assays have been developed and have remained the gold standard for detection. However, clinical significance of the presence of OXA-48 like genes cannot be inferred from genotypic testing³³ and thus, the detection of the carbapenemase that is actually expressed would provide important information. Moreover, for the PCR-methods specific primers are needed, requiring a priori knowledge that may even become problematic in case specific mutations occur in the corresponding target sequences.

Our method (total analytical analysis time of 60 min, separation window of 30 min) enables the detection of carbapenemase in all tested colony samples with the sequence coverage between 21 and 38%. This is lower than the maximum sequence coverage that we have obtained with a purified beta-lactamase (68%), but sufficient for a confident identification and (depending on the beta-lactamase) assignment of the beta-lactamase class or even species discrimination. The diagnostic timeframe for our method in the current setting would be around 12 to 16 hours including tryptic digestion, which is longer than PCR. However, having established the analytical platform to identify carbapenemases now permits improvement of the digestion protocol, for example using microwave-assisted protein digestion methods which could potentially bring back the overall analysis time to 5-6 hrs¹⁰.

As a result of the direct detection of the resistance linked enzymes, our CE-ESI-M/MS approach therefore offers an advantage over the methods described above. In every isolate analysed, the carbapenemase class as molecularly identified by PCR was also observed in CE-ESI-MS/

MS analysis and was successful in different Gram-negative species. Since the method does not provide full sequence coverage, it will not be able to identify all individual β-lactamases, much as molecular techniques would do without additional sequencing steps. However, with the sequence coverage observed, the method still provides at least carbapenemase class identification. Moreover, the unbiased nature of the approach allows for the detection of a variety of β-lactamase enzymes without a requirement of pre-existing sample

knowledge. Thus, in addition to the carbapenemases confirmed through PCR, a number of extended spectrum β-lactamases were identified in the present study. The strains had not been previously tested for these other genes. The multi-resistant character of the isolates is not surprising; not only is multi-resistance widespread, it has also been documented that carbapenem resistance is commonly observed in species that already harbour resistance to other β-lactam based antibiotics through ESBLs⁸.

Phenotypical tests such as E-tests can elucidate the degree of functional resistance per antibiotic³⁴. The meropenem MIC concentration in the isolates selected for the current study ranged from 0.5 to over 32 mg/L, indicating a varying degree of susceptibility across the different isolates. The analysis by CE-ESI-MS/MS seemed to be independent of this however, performing similarly for highly resistant and still partially susceptible cells. The MALDI-TOF MS ertapenem breakdown assay, on the contrary, only correctly identified 3 out of 17 OXA- 48 producers according to our definition, while KPC was easily detected with this method.

Besides mechanistic explanations, such as kinetic differences between KPC and OXA-48 or availability of the ertapenem to the respective carbapenemase pool, the dependence of this method on the complete disappearance of product peaks in the mass spectrum also played a role. Very recently, a novel method to detect carbapenem breakdown with MALDI- ToF MS, using ertapenem impregnated disks, was described, showing a good performance, also with OXA-48 positive cells³⁵. One potential problem with MALDI-TOF MS analysis of small molecules is related to the influence of the matrix effect. Therefore, we anticipate that CE- ESI-MS can potentially aid in this type of analysis, not only because of the elimination of the matrix effect, but also due to the separation of the substrate and products, resulting in more accurate detection and quantification capabilities. For similar reasons, several novel methods have recently been developed which allow a quantitative assessment of the carbapenem breakdown using single reaction monitoring on LC-ESI-MS/MS platforms ^36,37. Although not the major aim of the current experiments, our non-beta-lactamase proteomics data obviously also reveals the bacterial species (K. pneumoniae, E. coli or E. cloacae) harbouring the resistance phenotype. The total performance across an analysis generally results in 300-500 unique peptide identifications which lead to 100-200 protein identifications from as little as 10 ng of a tryptic digest. In other words, the analysis can combine species identification, similar as MALDI-ToF MS based approaches³⁸, but in addition reveals specific protein IDs and combines it with beta-lactamase resistance classification.

Obviously, more in-depth analysis, for example by using longer capillaries, capillaries with modified surface and/or multi-dimensional separations, may reveal other proteins involved in the resistance phenotype, such as proteins involved in drug uptake and excretion. However, such in-depth analysis is non-compatible with the throughput and speed of identification which the colony based protocol is demanding. Thus, as a next step, we are planning to

Chapter 2 perform similar analyses in samples taken directly from blood cultures which would be the

next step in speeding up the time till detection of the carbapenemase.

Conclusions

The growing incidence of β-lactam resistance asks for rapid, straightforward and clear identification of individual β-lactamases in clinical isolates to allow appropriate intervention strategies related to treatment and control of further spread. This study presents a novel CE- ESI-MS/MS workflow for the sensitive and unambiguous bottom-up proteomics identification of a OXA-48 and KPC in various Gram-negative species, even from single colonies. Additionally, due to the unbiased nature of the method, a number of non-targeted extended-spectrum β-lactamases were also identified, demonstrating the ability of the system to successfully assess multi-drug resistant bacteria. As such, it presents CE-ESI-MS as a potential next generation mass spectrometry platform for both species as well as resistance identification within the clinical microbiology laboratory.

Acknowledgements

This research was financially supported by the Netherlands organisation of scientific research (NWO, ZonMW grant number 50-51700-98-142). The authors would like to thank Ms.

S. Paltansing and Ms. M. Kraakman for technical assistance.

References

1 Boucher, H. W.; Talbot, G. H.; Benjamin, D. K., Jr.; Bradley, J.; Guidos, R. J.; Jones, R. N.; Murray, B. E.; Bonomo, R. A.; Gilbert, D.; Infect Dis Soc, A. Clin. Infect. Dis. 2013, 56, 1685-1694.

2 Huttner, A.; Harbarth, S.; Carlet, J.; Cosgrove, S.; Goossens, H.; Holmes, A.; Jarlier, V.; Voss, A.; Pittet, D.

Antimicrob Resist Infect Control 2013, 2, 31.

3 Yocum, R. R.; Rasmussen, J. R.; Strominger, J. L. J. Biol. Chem. 1980, 255, 3977-3986.

4 Poole, K. Cell. Mol. Life Sci. 2004, 61, 2200-2223.

5 Sacha, P.; Ostas, A.; Jaworowska, J.; Wieczorek, P.; Ojdana, D.; Ratajczak, J.; Tryniszewska, E. Folia Histochem Cytobiol 2009, 47, 537-543.

6 Nordmann, P.; Naas, T.; Poirel, L. Emerg. Infect. Dis. 2011, 17, 1791-1798.

7 Canton, R.; Akova, M.; Carmeli, Y.; Giske, C. G.; Glupczynski, Y.; Gniadkowski, M.; Livermore, D. M.; Miriagou, V.;

Naas, T.; Rossolini, G. M.; Samuelsen, O.; Seifert, H.; Woodford, N.; Nordmann, P.; European Network, C. Clin.

Microbiol. Infec. 2012, 18, 413-431.

8 Potron, A.; Poirel, L.; Rondinaud, E.; Nordmann, P. Euro Surveill. 2013, 18, 9-22.

9 Fournier, P.-E.; Drancourt, M.; Colson, P.; Rolain, J.-M.; La Scola, B.; Raoult, D. Nat. Rev. Microbiol. 2013, 11, 574-585.

10 Chang, C.-J.; Lin, J.-H.; Chang, K.-C.; Lai, M.-J.; Rohini, R.; Hu, A. Anal. Chem. 2013, 85, 2802-2808.

11 Faserl, K.; Sarg, B.; Kremser, L.; Lindner, H. Anal. Chem. 2011, 83, 7297-7305.

12 Schmidt, A.; Karas, M.; Dulcks, T. J. Am. Soc. Mass. Spectrom. 2003, 14, 492-500.

13 Marginean, I.; Kelly, R. T.; Prior, D. C.; LaMarche, B. L.; Tang, K. Q.; Smith, R. D. Anal. Chem. 2008, 80, 6573-6579.

14 Heemskerk, A. A. M.; Busnel, J. M.; Schoenmaker, B.; Derks, R. J. E.; Klychnikov, O.; Hensbergen, P. J.; Deelder, A. M.; Mayboroda, O. A. Anal. Chem. 2012, 84, 4552-4559.

15 Heemskerk, A. A.; Deelder, A. M.; Mayboroda, O. A. Mass Spectrom. Rev. 2014, In press.

16 Woodford, N.; Wareham, D. W.; Guerra, B.; Teale, C. J. Antimicrob. Chemother. 2014, 69, 287-291.

17 Lee, K.; Chong, Y.; Shin, H. B.; Kim, Y. A.; Yong, D.; Yum, J. H. Clin. Microbiol. Infect. 2001, 7, 88-91.

18 Sparbier, K.; Schubert, S.; Weller, U.; Boogen, C.; Kostrzewa, M. J. Clin. Microbiol. 2012, 50, 927-937.

19 Burckhardt, I.; Zimmermann, S. J. Clin. Microbiol. 2011, 49, 3321-3324.

20 Wang, H. X.; Qian, W. J.; Mottaz, H. M.; Clauss, T. R. W.; Anderson, D. J.; Moore, R. J.; Camp, D. G.; Khan, A. H.;

Sforza, D. M.; Pallavicini, M.; Smith, D. J.; Smith, R. D. J. Proteome Res. 2005, 4, 2397-2403.

21 Moini, M. Anal. Chem. 2007, 79, 4241-4246.

22 de Souza, G. A.; Arntzen, M. O.; Wiker, H. G. Bioinformatics 2010, 26, 698-699.

23 Ewing, A. G.; Wallingford, R. A.; Olefirowicz, T. M. Anal. Chem. 1989, 61, 292A-303A.

24 Olivares, J. A.; Nguyen, N. T.; Yonker, C. R.; Smith, R. D. Anal. Chem. 1987, 59, 1230-1232.

25 Lee, E. D.; Muck, W.; Henion, J. D.; Covey, T. R. J. Chromatogr. 1988, 458, 313-321.

26 Kelly, R. T.; Page, J. S.; Zhao, R.; Qian, W. J.; Mottaz, H. M.; Tang, K.; Smith, R. D. Anal. Chem. 2008, 80, 143-149.

27 Busnel, J. M.; Schoenmaker, B.; Ramautar, R.; Carrasco-Pancorbo, A.; Ratnayake, C.; Feitelson, J. S.; Chapman, J. D.; Deelder, A. M.; Mayboroda, O. A. Anal. Chem. 2010, 82, 9476-9483.

28 Maxwell, E. J.; Zhong, X.; Zhang, H.; van Zeijl, N.; Chen, D. D. Electrophoresis 2010, 31, 1130-1137.

29 Wojcik, R.; Dada, O. O.; Sadilek, M.; Dovichi, N. J. Rapid Commun. Mass Spectrom. 2010, 24, 2554-2560.

30 Sun, L. L.; Zhu, G. J.; Yan, X. J.; Dovichi, N. J. Curr. Opin. Chem. Biol. 2013, 17, 795-800.

31 Munoz-Price, L. S.; Poirel, L.; Bonomo, R. A.; Schwaber, M. J.; Daikos, G. L.; Cormican, M.; Cornaglia, G.; Garau, J.; Gniadkowski, M.; Hayden, M. K.; Kumarasamy, K.; Livermore, D. M.; Maya, J. J.; Nordmann, P.; Patel, J. B.;

Paterson, D. L.; Pitout, J.; Villegas, M. V.; Wang, H.; Woodford, N.; Quinn, J. P. Lancet Infect Dis 2013, 13, 785-796.

32 Nordmann, P.; Girlich, D.; Poirel, L. J. Clin. Microbiol. 2012, 50, 2761-2766.

33 Nordmann, P.; Poirel, L. J. Antimicrob. Chemother. 2013, 68, 487-489.

34 Pulido, M. R.; Garcia-Quintanilla, M.; Martin-Pena, R.; Cisneros, J. M.; McConnell, M. J. J. Antimicrob. Chemother.

2013, 68, 2710-2717.

35 Vogne, C.; Prod’Hom, G.; Jaton, K.; Decosterd, L.; Greub, G. Clin. Microbiol. Infect. 2014, In press.

Chapter 2

36 Wang, M.; Shen, Y.; Turko, I. V.; Nelson, D. C.; Li, S. Anal. Chem. 2013, 85, 11014-11019.

37 Peaper, D. R.; Kulkarni, M. V.; Tichy, A. N.; Jarvis, M.; Murray, T. S.; Hodsdon, M. E. Bioanalysis 2013, 5, 147-157.

38 Drake, R. R.; Boggs, S. R.; Drake, S. K. J. Mass Spectrom. 2011, 46, 1223-1232.